0

Is it possible to have mdadm automatically stop the array if any disk-failure is detected to limit data corruption?

Or better yet: Could it change the array to read-only? If so, would Linux handle it ok if one of its disks all-of-a-sudden went read-only?

Markus A.
  • 419
  • It already does this if there are not enough disks to continue. If there are enough disks to continue, why stop? – Michael Hampton Feb 06 '17 at 21:16
  • @MichaelHampton The problem I had is that 3 disks out of a RAID 6 had an intermittent failure a few hours apart. So it removed the 1st one and kept writing to the array, then the 2nd one and kept writing, and eventually the third one and only then it killed the array. So, when I tried to reassemble the array after fixing the cabling issue that caused the problem, I had 3 "out of date" disks and had to force the array back online. I would much rather have 1 "out of date" disk that can then be rebuilt with certainty... – Markus A. Feb 06 '17 at 22:12

0 Answers0